Successor Features in Deep Reinforcement Learning and Transfer Learning

This project takes part into the master 'MVA'. It is based on the following papers :

• Barreto, A., Dabney, W., Munos, R., Hunt, J. J., Schaul, T., Silver, D., & van Hasselt, H. P. (2017). Successor features for transfer in reinforcement learning. In Advances in Neural Information Processing Systems (pp. 4058-4068).

• Kulkarni, T. D., Saeedi, A., Gautam, S., & Gershman, S. J. (2016). Deep successor reinforcement learning. arXiv preprint arXiv:1606.02396.

• Lehnert, Lucas, Stefanie Tellex, and Michael L.Littman. Advantages and Limitations of using Successor Features for Transfer in Reinforcement Learning. arXiv preprint arXiv:1708.00102 (2017).

Summary

Implementation of the Puddle World described in Barreto et al. (2017) and comparing classical reinforcement learning algorithm to transfer learning with successor feature algorithm. See report.pdf for more description.

Review papers using similar algorithms in deep reinforcement learning.